AI/LLMs Efficiency - When the AI turns on Humanity

[Umbran]

I can't see anything they get out of scaring the public into being even more anti-AI than they already are. This will only hurt them, not help.

It is unclear how much they care about the opinions of the anti-AI public. They need tons of money to keep afloat, and that's not coming from the general public.

It is pretty clear they care about their stock prices and investor opinion. And investors will look at that report and say, "AI is badass!" They see power and will invest in power.

 

[UngainlyTitan]

I wonder what these companies have to gain by claiming every other month that they built SkyNet.

It does help to convince people that the IPO might be worth it despite the companies being a money furnace.

 

[Justice and Rule]

It is unclear how much they care about the opinions of the anti-AI public. They need tons of money to keep afloat, and that's not coming from the general public.

It is pretty clear they care about their stock prices and investor opinion. And investors will look at that report and say, "AI is badass!" They see power and will invest in power.

This is exactly it. Lying about the capabilities like it's ability to hack doesn't really create outrage, but what it does do is create a whole bunch of hype about what your thing can do. AI can autonomously hack other servers? CEOs who don't know anything are going to be like "We should get an AI to protect us against that!" and government officials look at it as "Oh man, we need this to make sure we don't lose ground to <insert enemy here>!" It's why the extreme AI Doomers who think that Skynet is coming are really just building more hype for the idea than anything, trying to build it up as a existential threat that it isn't: saying that AI has that sort of power only allows the marketing hype teams to say "Sure, it could do that, but we're responsible!", taking a completely unproven claim, validating it for their product, and using it to build up the power of it.

 

[Umbran]

This is exactly it. Lying about the capabilities like it's ability to hack doesn't really create outrage, but what it does do is create a whole bunch of hype about what your thing can do.

"Why would they do X?" Rule #1 - Follow the money.

 

[Justice and Rule]

"Why would they do X?" Rule #1 - Follow the money.

Hey, funny that: OpenAI has their own AI Defensive Software. They mentioned it at the end of the release on North Korea using ChatGPT to hack things ...